Auxiliary Material for
Real-time inversions for finite fault slip models and rupture geometry based on high-rate GPS data.
Sarah E. Minson1, Jessica R. Murray2, John O. Langbein2, and Joan S. Gomberg1
(1USGS Earthquake Science Center, Seattle, Washington, USA,
2
USGS Earthquake Science Center, Menlo Park, California, USA.)
Journal of Geophysical Research, Solid Earth, 2013
Introduction
This supplementary material contains data set contains movies depicting the finite fault slip model
for each of the three test earthquakes (2011 Tohoku-oki, 2003 Tokahi-oki, Hayward fault scenario)
using each of three different data processing schemes (raw data, data smoothed with an
exponentially weighted moving average (EWMA) with a decay constant of 0.1, and data smoothed
with an EWMA with a decay constant of 0.01). The first supplementary figure illustrates the
sensitivity of the inversion to the choice of a priori covariance on the slip parameters. Additional
supplementary figures show the inversion results that were animated in the movies in fault-view so
that the distribution of slip on the fault may be more clearly seen. The final nine supplemental
figures are the same as Supplemental figures 5-13 are the same as Figures 13-21 in the main paper
except the black and red vectors show the observed and predicted vertical surface displacements
and the gray vectors show the slip direction on each patch with length scaled by the slip on that
patch.
1. ms01.avi (48MB AVI file) Inversion of Tohoku-oki 1 Hz GPS data using offsets estimated
from raw time series.
2. ms02.avi (44MB AVI file) Inversion of Tohoku-oki 1 Hz GPS data using offsets estimated
from time series smoothed using an exponentially weighted moving average (EWMA) with a
decay constant of 0.1.
3. ms03.avi (39MB AVI file) Inversion of Tohoku-oki 1 Hz GPS data using offsets estimated
from time series smoothed using an EWMA with a decay constant of 0.01.
4. ms04.avi (24MB AVI file) Inversion of Tokachi-oki 1 Hz GPS data using offsets estimated
from raw time series.
5. ms05.avi (23MB AVI file) Inversion of Tokachi-oki 1 Hz GPS data using offsets estimated
from time series smoothed using an EWMA with a decay constant of 0.1.
6. ms06.avi (21MB AVI file) Inversion of Tokachi-oki 1 Hz GPS data using offsets estimated
from time series smoothed using an EWMA with a decay constant of 0.01.
7. ms07.avi (13MB AVI file) Inversion of Hayward fault simulated 1 Hz GPS data using offsets
estimated from raw time series.
8. ms08.avi (12MB AVI file) Inversion of Hayward fault simulated 1 Hz GPS data using offsets
estimated from time series smoothed using an EWMA with a decay constant of 0.1.
9. ms09.avi (11MB AVI file) Inversion of Hayward fault simulated 1 Hz GPS data using offsets
estimated from time series smoothed using an EWMA with a decay constant of 0.01.
10. fs01.pdf (Supplementary figure 1) Effects on inversion results for slip of varying V. The mean
posterior estimate for the strike-slip (left) and dip-slip (right) components of motion are
shown for a simulated Mw 9 Tohoku-oki-type earthquake. For reference, the input slip
model is shown in the top row.
11. fs02.pdf (Supplementary figure 2) Detail of Tohoku-oki earthquake slip models plotted in
Figures 13-15.
12. fs03.pdf (Supplementary figure 3) Detail of Tokachi-oki earthquake slip models plotted in
Figures 16-18.
13. fs04.pdf (Supplementary figure 4) Detail of Hayward fault simulation slip models plotted in
Figures 19-21.
14. fs05.pdf (Supplementary figure 5) Result of semi-analytical Bayesian real-time finite fault
inversion for the Tohoku-oki earthquake at 60 seconds after the origin time. Same as
Figure 13 except black and red vectors show observed and predicted GPS displacements,
respectively, and gray vectors are direction of slip on each patch scaled by slip magnitude.
15. fs06.pdf (Supplementary figure 6) Result of semi-analytical Bayesian real-time finite fault
inversion for the Tohoku-oki earthquake at 120 seconds after the origin time. Same as
Figure 14 except black and red vectors show observed and predicted GPS displacements,
respectively, and gray vectors are direction of slip on each patch scaled by slip magnitude.
16. fs07.pdf (Supplementary figure 7) Result of semi-analytical Bayesian real-time finite fault
inversion for the Tohoku-oki earthquake at 180 seconds after the origin time. Same as
Figure 15 except black and red vectors show observed and predicted GPS displacements,
respectively, and gray vectors are direction of slip on each patch scaled by slip magnitude.
17. fs08.pdf (Supplementary figure 8) Result of semi-analytical Bayesian real-time finite fault
inversion for the Tokachi-oki earthquake at 30 seconds after the origin time. Same as
Figure 16 except black and red vectors show observed and predicted GPS displacements,
respectively, and gray vectors are direction of slip on each patch scaled by slip magnitude.
18. fs09.pdf (Supplementary figure 9) Result of semi-analytical Bayesian real-time finite fault
inversion for the Tokachi-oki earthquake at 60 seconds after the origin time. Same as
Figure 17 except black and red vectors show observed and predicted GPS displacements,
respectively, and gray vectors are direction of slip on each patch scaled by slip magnitude.
19. fs10.pdf (Supplementary figure 10) Result of semi-analytical Bayesian real-time finite fault
inversion for the Tokachi-oki earthquake at 90 seconds after the origin time. Same as
Figure 18 except black and red vectors show observed and predicted GPS displacements,
respectively, and a gray reference vector shows the direction of slip on the patch with
maximum slip.
20. fs11.pdf (Supplementary figure 11) Result of semi-analytical Bayesian real-time finite fault
inversion for the Hayward fault simulation at 20 seconds after the origin time. Same as
Figure 19 except black and red vectors show observed and predicted GPS displacements,
respectively, and gray vectors are direction of slip on each patch scaled by slip magnitude.
21. fs12.pdf (Supplementary figure 12) Result of semi-analytical Bayesian real-time finite fault
inversion for the Hayward fault simulation at 40 seconds after the origin time. Same as
Figure 20 except black and red vectors show observed and predicted GPS displacements,
respectively, and a gray reference vector shows the direction of slip on the patch with
maximum slip.
22. fs13.pdf (Supplementary figure 13) Result of semi-analytical Bayesian real-time finite fault
inversion for the Hayward fault simulation at 60 seconds after the origin time. Same as
Figure 21 except black and red vectors show observed and predicted GPS displacements,
respectively, and a gray reference vector shows the direction of slip on the patch with
maximum slip.